Freshness measurement of eggs using near infrared (NIR) spectroscopy and multivariate data analysis

Near infrared (NIR) spectroscopy combined with multivariate analysis were attempted to determine freshness of eggs. Independent component analysis (ICA) and principle component analysis (PCA) algorithms were performed comparatively to extract effective features from the original data. Artificial neural network combined with genetic algorithms (GA-ANN) were employed to calibrate regression model. Some parameters of GA-ANN model were optimized by cross-validation in building models. The performance of the final optimized model was evaluated according to the root mean square error of prediction (RMSEP) and the correlation coefficient (R) in the prediction set. The optimal performance was obtained when 7 ICs from ICA were used in GA-ANN model. It was achieved with RMSEP = 2.443 and R = 0.879. This work shows that NIR spectroscopy combined with multivariate calibration has significant potential in the analysis of freshness of eggs. Industrial relevance: Freshness makes a major contribution to the quality of egg and egg products, because consumers may perceive variability in freshness as lack of quality. Egg freshness grading was mostly relied on storage time. However, based on individual differences of eggs, freshness of them fluctuated. Consumers might buy un-fresh eggs from the market. This work presents a non-destructive method for the measurement of egg freshness, and builds a robust calibration model to improve the prediction ability. The research data presents a potential way for fast, non-destructive and automatic measurement of freshness in egg industry. (C) 2011 Elsevier Ltd. All rights reserved.